Automated Mapping of Sensorimotor Network for Resting State fMRI Data with Seed-Based Correlation Analysis Bruno Goulart de Oliveira, José Osmar Alves Filho, Nathalia Bianchini Esper, Dario Francisco Guimaraes de Azevedo, and Alexandre R. Franco Abstract An algorithm for automated placement of regions of interest (ROI) in Seed Based Correlation (SBC) data analysis for resting-state functional Magnetic Resonance Imaging (rs-fMRI) is presented in this paper. The sensorimotor network was used for testing and validation. Most of the available literature shows the use of manual seed selection in order to find the Resting-State Networks (RSNs). Typically, a seed is placed in the most preserved side of brain and its functional connectivity (correlation) with the contra-lateral hemisphere allows the identifica- tion of the network within the lesioned side of the brain. The manual placement of the seeds is usually a laborious task and prone to human error. The developed algorithm was based on the automated spatial registration of an atlas to the space of the patient’s brain: Anatomical (Har- vardOxford) and functional (Brodmann Areas) atlas. Regions of interest representing the sensorimotor net- works were used as seeds. FMRI data from 8 healthy volunteers were used to assess its validation. These data included a finger-tapping task and a resting-state protocol. The extracted sensorimotor RSNs derived from the automated procedure were compared to the task-based fMRI maps and RSNs extracted from SBC with manual ROI placement. Preliminary results show a good level of similarity between seed-based and task-based motor network maps, except in one case in which the patterns did not match. This technique shows potential to be used in clinical application due to the automated nature of the data processing as well as the ease for patients to perform the exam. Keywords Presurgical planning Á Resting-state fMRI Á Neuroimage preprocessing 1 Introduction An intervention such as the surgical resection of brain tissue is sometimes needed due to the presence tumor cells. When brain surgery is being planned, the eloquent cortex needs to be mapped. It contains the most important region of the brain and its corresponding functions. These areas of the brain need to be preserved so the patient could have a normal life after surgery. These functions include: movement, sensitiv- ity, vision and language [1]. Accurate localization of this areas helps to optimize resection and minimize postoperative neurological deficits [2]. The current gold technique for eloquent cortex map is Electro-Cortical Stimulation (ECS) [3]. Although very reli- able and accurate, ECS has some limitations: (1) it is highly invasive; (2) it cannot be used at the presurgical planning stage but only intra-operatively; (3) it is limited in access by the surgical operculum; and (4) it may occasionally yield inconclusive results due to anesthesiologic issues [1, 4–8]. To overcome some limitation of ECS, task-based func- tional Magnetic Resonance Image (tb-fMRI) mapping may be used. It has proven useful and has shown good corre- spondence with ECS for motor areas [7, 9] and Wada testing for language lateralization [9–11]. Tb-fMRI measures B. G. de Oliveira Á D. F. G. de Azevedo School of Technology, PUCRS, Porto Alegre, RS 90619-900, Brazil e-mail: bruno.goulart@acad.pucrs.br J. Osmar Alves Filho New York University, New York City, NY 10016, USA N. B. Esper School of Medicine, PUCRS, Porto Alegre, RS 90619-900, Brazil N. B. Esper Brain Institute of Rio Grande do Sul, PUCRS, Porto Alegre, RS 90619-900, Brazil A. R. Franco (&) Nathan Kline Institute, Orangeburg, NY 10962, USA e-mail: alexandre.franco@childmind.org A. R. Franco Child Mind Institute, New York City, NY 100022, USA © Springer Nature Singapore Pte Ltd. 2019 R. Costa-Felix et al. (eds.), XXVI Brazilian Congress on Biomedical Engineering, IFMBE Proceedings 70/2, https://doi.org/10.1007/978-981-13-2517-5_81 537